As a non-relational (NoSQL) database of distributed storage, a database of a key-value pair (KeyValue) type (which is hereinafter referred to as a KeyValue database) has high scalability and high reliability and has been widely applied in more systems. Data in many KeyValue databases is stored in units of table. Each table includes multiple rows of data; each row of data is uniquely identified by using a row key (RowKey), and each row of data includes multiple column qualifiers; and each column qualifier is corresponding to a piece of KeyValue data, and each column qualifier has a data type and a timestamp. The data type includes Put (newly-added), Delete (delete), and the like. A Put type is used to indicate that a qualifier is a newly-added qualifier, and a Delete type is used to indicate that a qualifier is used to delete an old qualifier. The timestamp is used to indicate generating time of each qualifier. In a KeyValue database, multi-version storage of data is implemented by using a timestamp, that is, for data with a same RowKey value, new data and old data are differentiated by using a timestamp. When the KeyValue database includes multiple versions of data, an old version of data is overwritten by a new version of data, and when reading data, a user directly reads the new version of data.
A mark-as-delete technique is used for updating data in the KeyValue database, and KeyValue data is deleted or imported piece by piece. In an actual application, full updating often needs to be performed on all data of table data, that is, all existing data in a table needs to be cleared and new data is imported into the table. However, an existing KeyValue database does not support one-time full updating. If table data is deleted and imported piece by piece, a process of updating all data in a table takes a long time, and the updating process lacks atomicity, which affects quality of a data reading service provided by the table. Therefore, it is of great significance to provide a table data full updating method applicable to the KeyValue database.